Simplex tree expansion_with_blockers python interface

[VincentRouvreau]

This is a prototype based on https://github.com/cython/cython/tree/master/Demos/callback

from gudhi import SimplexTree
st=SimplexTree()

def blocker(simplex):
    try:
        # Block all simplices that countains vertex 6
        simplex.index(6)
        print(simplex, ' is blocked')
        return True
    except ValueError:
        print(simplex, ' is accepted')
        print(simplex, " = ", st.filtration(simplex))
        st.assign_filtration(simplex, 12.)
        return False

st.insert([0,1],0)
st.insert([0,2],1)
st.insert([0,3],2)
st.insert([1,2],3)
st.insert([1,3],4)
st.insert([2,3],5)
st.insert([2,4],6)
st.insert([3,6],7)
st.insert([4,5],8)
st.insert([4,6],9)
st.insert([5,6],10)
st.insert([6],10)

st.expansion_with_blocker(2, blocker)
print(list(st.get_filtration()))

[6, 5, 4]  is blocked
[3, 2, 1]  is accepted
[3, 2, 1]  =  5.0
[3, 2, 0]  is accepted
[3, 2, 0]  =  5.0
[2, 1, 0]  is accepted
[2, 1, 0]  =  3.0
[3, 1, 0]  is accepted
[3, 1, 0]  =  4.0

[([0], 0.0), ([1], 0.0), ([0, 1], 0.0), ([2], 1.0), ([0, 2], 1.0), ([3], 2.0), ([0, 3], 2.0), ([1, 2], 3.0), ([1, 3], 4.0), ([2, 3], 5.0), ([4], 6.0), ([2, 4], 6.0), ([6], 7.0), ([3, 6], 7.0), ([5], 8.0), ([4, 5], 8.0), ([4, 6], 9.0), ([5, 6], 10.0), ([0, 1, 2], 12.0), ([0, 1, 3], 12.0), ([0, 2, 3], 12.0), ([1, 2, 3], 12.0)]

[mglisse]

Please don't use blocker_func both as a type here and as the name of a variable in another file. Maybe add _t for the type?

[VincentRouvreau]

Renamed on f4f5992

[mglisse]

The C++ version allows that, it would be good to understand what is going wrong here.

[mglisse]

I guess this paragraph is now wrong?
The C++ doc has a warning about phantom simplices that should also be in the python doc.

[VincentRouvreau]

The note was removed on 924302f
Add a warning about phantom vertices in f6a4524

[mglisse]

:type blocker_func: Callable[[List[int]], bool]
according to https://docs.python.org/3/library/typing.html#callable

[VincentRouvreau]

Done on f4f5992

[mglisse]

Is it because simplex->second.assign_children(new_sib); happens too late, would moving it earlier (and cancelling it in the case where we erase everything) help?
(looking at this code, the quadratic erase loop doesn't look so good, but that's just a performance question)

[VincentRouvreau]

Is it because simplex->second.assign_children(new_sib); happens too late, would moving it earlier (and cancelling it in the case where we erase everything) help? (looking at this code, the quadratic erase loop doesn't look so good, but that's just a performance question)

Nice ! Thanks ! It has been done on d114eaf